Analysis of Oxide Scales Formed in the Naphthenic Acid Corrosion of

Jul 8, 2016 - ... Ohio University, 342 West State Street, Athens, Ohio 45701, United States ... In depth analysis identified magnetite in the sections...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/EF

Analysis of Oxide Scales Formed in the Naphthenic Acid Corrosion of Carbon Steel Peng Jin,* Gheorghe Bota, Winston Robbins, and Srdjan Nesic Institute for Corrosion and Multiphase Technology, Ohio University, 342 West State Street, Athens, Ohio 45701, United States ABSTRACT: Naphthenic acid corrosion of steel is a major challenge in oil refineries. Iron sulfide scales, formed by corrosion due to sulfur compounds found in the crude oil, have shown unpredictable behavior when it comes to their protectiveness. Recent results show that simultaneous formation of an iron oxide scale formed by corrosion due to naphthenic acids may explain part of the variability and contribute to corrosion resistance of the scale formed on 5Cr steel. In depth analysis identified magnetite in the sections of the scale just adjacent to the metal surface. Currently reported research focuses on investigating conditions that lead to the formation of the iron oxide layer on carbon steel. A comparison of different oxide layers formed on carbon steel from a pure model acid (palmitic acid) and a commercial naphthenic acid mixture derived from petroleum shows different behavior when it comes to corrosion protection. The two acids also show different behavior in the presence of sulfur containing compounds (a model compound and native sulfur compounds found in a heavy lube basestock). These results suggest that the molecular structure of the acid is an important factor in the formation and the protectiveness of an iron oxide scale.

1. INTRODUCTION The carboxylic acid found in petroleum is called “naphthenic acid”.1 Processing low-price crude oil with high content of naphthenic acid can increase the profit of a medium-size crude oil refinery by over ten million dollars per year.2 But naphthenic acid is one of the major causes of high temperature (220−400 °C) corrosion in refinery distillation towers. Naphthenic acid corrosion (NAC) of steel is widely attributed to the formation of soluble iron naphthenates, leaving the steel surface exposed to further attack.3 In contrast, reactive sulfur compounds corrode the steel in the same temperature range (sulfidation) and form insoluble iron sulfide scales.4−6 The iron sulfide scales have been researched extensively in the past, given that they have been thought to lead to a decrease in corrosion rates. Because NAC and sulfidation inevitably occur simultaneously, chemical/corrosion engineers face a major challenge in predicting and mitigating high temperature corrosion. Specifically, protective effects of scales have proven to be difficult to predict, not only because of the wide diversity of sulfur compounds found in crude oil but also because of their interaction with naphthenic acids. The corrosion process is often described by three generic free radical reactions that occur at high temperatures, in the absence of water: Fe(s) + 2RCOOH → Fe(RCOO)2 + H 2

(R1)

Fe(s) + H 2S → FeS(s) + H 2

(R2)

Fe(RCOO)2 + H 2S ⇌ FeS(s) + 2RCOOH

(R3)

Some sulfur compounds found in petroleum may react directly with a steel surface, or may first decompose to form hydrogen sulfide, which then reacts with the iron in the steel, as shown by reaction R2, a process often referred to as simply “sulfidation”. In either case insoluble iron sulfide is formed, which remains as a deposit on the steel surface in the form of a scale. Again it is assumed that there is no buildup of hydrogen concentration in the oil and therefore the reverse reaction can be ignored. A possible interaction between the two corrosion pathways (NAC and sulfidation) is indicated by reaction R3. It is suggested that hydrogen sulfide may react rapidly with the iron naphthenate in the oil phase to form insoluble iron sulfide and “regenerate” the naphthenic acid. Alternatively, in solutions with high concentrations of naphthenic acids the reverse reaction will dominate and the iron sulfide scale may be redissolved. This last possibility leads to a concern that in the presence of naphthenic acids, any protectiveness offered by the iron sulfide scale may be compromised leading to elevated corrosion rates. The reaction kinetics are dependent on characteristics of the various naphthenic acids and sulfur compounds involved in the process. Crude oil is one of the most complex fluids found in nature, due to the wide variability in its geochemical origin; within any given crude oil sample more than 105 individual components are routinely detected.7 Therefore, the naphthenic acids and sulfur compounds present in the crude oil vary widely. Nowadays, the number of different naphthenic acids identified in heavy crude oils has grown far beyond 3000, detected 15 years ago (where it was assumed that R = CxHy, see reaction R1).8 It is now recognized that there may be many

Reaction R1 indicates that naphthenic acids oxidize iron to form oil soluble salts (iron naphthenates) and hydrogen gas. In this reaction, R represents the hydrocarbon portion of the naphthenic acid molecule and COOH is the corrosive carboxylic acid functional group. The reverse reaction is ignored here as it is assumed that there is no buildup of iron naphthenate and hydrogen concentration in the oil. © XXXX American Chemical Society

Received: May 3, 2016 Revised: June 9, 2016

A

DOI: 10.1021/acs.energyfuels.6b01066 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels Table 1. Chemical Composition of A106 Carbon Steel Specimen (wt %) C

Si

Mn

P

S

Cr

Ni

Mo

V

Cu

Fe

0.18

0.41

0.8

0.11

0.06

0.02

0.04

0.02

0.03

0.08

bal

heteroatom acids present as well (where R = CxHyNzSv).9−11 Even more different organo-sulfur compositions can be found in crude oils. Organo-sulfur compositions include nonreactive sulfur compounds (predominantly thiophenes) and reactive sulfur compounds with different molecular structure, functionality and reactivity, such as mercaptans (R−SH), aliphatic and alicyclic sulfides (R−S−R), and disulfides (R−S−S−R).12,13 Given the complexity of crude oils, it is not surprising that it is difficult to correlate any given crude oil composition with the observed corrosion attack. For practical purposes, studies of sulfidation and NAC were frequently related to the total sulfur content (%S) and the total naphthenic acid concentration, as seen in refinery streams, irrespective of the variability in their molecular structure. The %S is routinely determined by X-ray fluorescence (XRF) or other comparable methods. The naphthenic acid concentration is most commonly expressed as the total acid number (TAN), which represents the amount of KOH in milligrams needed to neutralize the naphthenic acid present in one gram of oil, obtained usually by titration. Sulfidation and NAC studies are difficult as it is hard to reproduce the real conditions encountered in refinery operations in laboratory experiments. High temperature, high velocity, high oil volume/metal area ratio, large time scale of corrosion effects such as growth of corrosion product scales, varying conditions, etc., all make it hard to do convincing laboratory studies. This is probably the reason that the important effect of oil composition has rarely been the focus of previous studies. Few publications deal with the effects of molecular weight and molecular structure of the corrosive species such as naphthenic acids and sulfur compounds.14,15 To avoid dealing with the complexity of real crude oils one approach is to work with representative model compounds. This would enable us to investigate the least understood and probably one of the most important factors: the interaction between sulfur compounds and naphthenic acids. One of the breakthrough discoveries made by the present authors was the identification of a thin protective oxide scale under the much thicker sulfide scale.13 Although oxygen in the corrosion product scale had been detected previously in studies of other groups, its presence was dismissed as the exposure to air contamination during or after the experiment.17 Previously, we reported that naphthenic acids were the source of oxygen in the formation of a protective oxide scale on the surface of chromium steel using a unique experiment protocol. In this protocol, a commercial mixture of naphthenic acids and a model sulfur compound were used to “pretreat” steel specimens in the form of rings to develop a scale and then the scale was attacked by a “challenge” solution containing an elevated concentration of naphthenic acids dissolved in a white mineral oil. This work led to the discovery of a thin oxide scale adjacent to the steel surface that appeared to contribute to the protectiveness against corrosion. This protective oxide scale, characterized as magnetite, has been proposed to arise from thermal decomposition of the iron naphthenate on 5Cr steel surfaces.16 Other sources of oxygen such as air and water were eliminated as possibilities, as described below. While the results relating to 5Cr steel have been reported elsewhere,16,38 current research focuses on experiments

involving carbon steel. In addition to the native naphthenic acids mixture used previously, this study compares it with the effect produced by a model fatty acid−palmitic acid. The effect of sulfur compound was previously studied by using a model compound: n-dodecyl sulfide (DDS) dissolved in white oil, whereas here it was compared with the attack caused by native sulfur compounds found in a Group I lube basestock (yellow oil).

2. EXPERIMENTAL SECTION 2.1. Experiment Materials. Rings made from A106 carbon steel (CS), 81.76 mm outer diameter, 70.43 mm inner diameter, and 5 mm thickness, were used as experiment specimens (for chemical composition of the steel see Table 1). Prior to the experiment, each specimen was polished with 400 and 600-grit silicon-carbide paper (SiC) under the flow of isopropanol in order to avoid oxidation and overheating. Then, the specimen was rinsed with toluene and acetone to clean the surface from any organic deposits and dried under the flow of nitrogen. The weight of the specimen was taken using an analytical balance with the accuracy of 0.1 mg. All experiments were performed by exposing three CS specimens to the corrosive hydrocarbon solution. Two specimens were used for weight loss measurements, and the third was used for microscopic examination. After the experiment, specimens for weight loss measurements were rinsed with toluene and acetone, dried, brushed to mechanically remove loose scale, exposed to the Clarke solution (ASTM G103),18,19 to dissolve the remaining scale, dried again, and finally reweighed. The corrosion rate was calculated on the basis of the weight loss, i.e., by subtracting the weight before and after the experiment. Specimens that were examined by microscopy were not subjected to the same cleaning procedure; rather, they were stored in a mineral oil to prevent degradation of the surface scale until time came for analysis, rinsed with toluene and acetone, dried, mounted, and examined using a microscope. The corrosion product scale formed on specimen surfaces was analyzed by a JEOL JSM-6390 scanning electron microscope (SEM). Some scales were analyzed by FEI Helios Nanolab 650 for focused ion beam (FIB) and Zeiss Libra 200EF transmission electron microscope (TEM). The crystal structure of scale was investigated by X-ray diffraction (XRD) on Bruker Discover D8 with a Co Kα X-ray tube. 2.2. Experiment Solutions. Recrystallized n-dodecyl sulfide (DDS) by Fisher Chemical was used as a model compound to represent the reactive sulfur compounds found in the crude oil. The mixture of naphthenic acids (TCI) by TCI America (Table 2) was

Table 2. Boiling Point Range of the Model Naphthenic Acid TCI (TAN 230) parameter

temperature (°C)

initial boiling point (IBP) 50% boiling point 80% boiling point final boiling point (FBP)

239 296 343 493

used as a model compound for naphthenic acids found in crude oil. Palmitic acid (PA) by Fisher Chemical, which is a simple fatty acid CH3(CH2)14COOH with carbon number close to the TCI, was also used as a model acid compound.7 These acids and/or DDS were dissolved into a paraffinic white oil solvent (Tufflo 6056, CITGO) in order to prepare the corrosive solutions (see Table 3 for properties of this solvent). In different experiments, acids were dissolved in a “yellow oil” solvent, which is a Group I lube basestock (America’s B

DOI: 10.1021/acs.energyfuels.6b01066 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

reactor and rotated. The cross section of the HVR reactor is shown in Figure 1. 2.4. Experimental Protocol. The pretreatment/corrosion challenge experiment protocol consisted of two consecutive steps: • the pretreatment step used to create the corrosion product scale in different solutions as listed in Table 5 and • the corrosion challenge step where the scale was exposed to the challenge solution. For the pretreatment in the stirred autoclave, the corrosion product scale formed on specimens that were fully immersed in 0.7 L corrosive solution. Before the starting of the pretreatment, the headspace of autoclave was purged with nitrogen gas to remove oxygen. During a 24 h pretreatment at 343 °C, the specimens were stagnant while mixing was done by an impeller rotating at 500 rpm. At the end of the pretreatment, the autoclave was cooled and the oil was drained, the specimens rinsed, and used for weight loss and microscopic analysis. Another parallel pretreatment experiment was run under same conditions, but specimens with their intact corrosion product scales were transferred to the HVR for the challenge step. In the corrosion challenge, pretreated specimens were mounted in the HVR, heated to 343 °C as hot Tufflo flowed through the reactor. The specimens were rotated at 2000 rpm (corresponding to a peripheral velocity of 8.56 m/s, Reynolds number of 1771, and wall shear stress of 74 Pa). The actual corrosion challenge experiment began when the input flow was switched from Tufflo to the challenge solution, and lasted for 24 h with the challenge solution flowing through the reactor at 7.5 cm3/min. The back-pressure of 150 psig was applied to suppress gas breakout. The corrosion challenge ended when the flow was switched back to Tufflo. The specimens were allowed to cool in Tufflo before being removed from the HVR for analysis. 2.5. Evaluation of Corrosion Rates. Time averaged corrosion rates of the specimens were calculated based on their weight loss during the experiment. For the pretreatment experiment conducted in the stirred autoclave, the corrosion rate was calculated as follows:

Table 3. Selected Physical and Chemical Properties of Model OilTufflo 6056 parameter

description

appearance color odor density (at 16 °C, kg/m3) flash point (°C) average molecular weight (g/mol) initial boling point (°C)

clear liquid colorless odorless 876 254 530 388

Core 600, Exxon Chemical) with no naphthenic acid and 0.25%S by weight (see Table 4 for its properties).

Table 4. Selected Physical and Chemical Properties of Model Oil−Yellow Oil (America’s Core 600) parameter

description

appearance color odor density (at 15 °C, kg/m3) flash point (°C)

clear liquid yellow odorless 879 270

There were a few different types of corrosive solutions used in the experiments, which were prepared as follows: (1) PA only solution that had no sulfur compounds with only palmitic acid dissolved in Tufflo. (2) TCI only solution that had no sulfur compounds with only TCI dissolved in Tufflo. (3) PA + DDS solution that had both acidic and sulfur compound obtained by dissolving PA and DDS in Tufflo. (4) “TCI + DDS” solution that had both naphthenic acids and sulfur compound obtained by dissolving TCI and DDS in Tufflo. (5) PA + Yellow oil solution that had both acidic and sulfur compounds obtained by dissolving PA in Yellow oil. (6) “TCI + Yellow oil” solution that had both naphthenic acids and sulfur compounds obtained by dissolving TCI in Yellow oil. (7) “Challenge solution” that had no sulfur compounds with only TCI dissolved in Tufflo. The TAN value and sulfur content of each corrosive solution are shown in Table 5.

CR pretreatment =

(1) where CRpretreatment is pretreatment corrosion rate [mm/y]; IW is initial weight of freshly polished specimen [g], FW is final weight of specimen after treatment with Clarke solution [g], ρsteel is density of specimen [g/cm3], As,pretreatment is area of specimen exposed to corrosive fluid during pretreatment [cm2], tpretreatment is duration of pretreatment [h]. In the corrosion challenge step, the freshly polished specimen was first pretreated in the autoclave followed by the corrosion challenge in the HVR. The challenge corrosion rate was calculated as follows:

Table 5. Properties of Corrosive Solutions solution

TAN

PA only TCI only PA + DDS TCI + DDS PA + Yellow oil TCI + Yellow oil challenge solution

0.5, 1, or 1.75 1.75 1.75 1.75 1.75 1.75 3.5

CR challenge =

%S 0 0 0.25 0.25 0.25 0.25 0

wt wt wt wt

(IW − FW) × 10 × 24 × 36 ρsteel A s,pretreatment t pretreatment

(IW − FW − WLpretreatment) ρsteel A s,challengetchallenge

× 10 × 24 × 36

(2) where CRchallenge is net corrosion rate from the corrosion challenge step (excluding the pretreatment step) [mm/y], IW is initial weight of freshly polished specimen [g], FW is final weight of specimen after treatment with Clarke solution [g], WLpretreatment is weight loss of specimen in the pretreatment step [g], ρsteel is density of specimen [g/ cm3], As,challenge is area of specimen exposed to corrosive fluid during challenge [cm2], tchallenge is duration of corrosion challenge [h].

% % % %

3. RESULTS AND DISCUSSION 3.1. Effect of Palmitic Acid Concentration on Formation of Oxide Scale. Corrosion rates measured for the CS specimens during the pretreatment step at different TAN (PA concentrations) are shown in Table 6. The thickness of the corrosion product scale was proportional to the PA concentration (TAN) as seen in the cross-section SEM images below, Figure 2a−c. The EDS analysis shows that the scale is

2.3. Experimental Equipment. The equipment described in our previous publications was used in the present corrosion experiments following the pretreatment/corrosion challenge protocol. A 1 L stirred autoclave was used for the pretreatment of CS specimens. The corrosion challenge was performed in a so-called high velocity rig (HVR) which is a flow-through rotating cylinder reactor. As depicted in our prior publication, the HVR was designed to create a high flow velocity and associated turbulence and shear stress.20 The CS specimens were mounted on a mandrel in the core of the HVR C

DOI: 10.1021/acs.energyfuels.6b01066 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

Figure 1. Cross-sectional view of HVR reactor.

Table 6. Pretreatment Corrosion Rates of CS Specimens in the PA Only Solution TAN of PA only solution

pretreatment corrosion rate (mm/y)

0.5 1 1.75

0.1 0.2 0.2

Table 7. Pretreatment and Challenge Corrosion Rates of CS Specimens Pretreated in PA only, PA + DDS, and PA + Yellow Oil

rich in oxygen (Figure 2a′−c′). The specimen with the thickest scale, exposed to the TAN 1.75 solution, was selected as a baseline for the subsequent experimentation. 3.2. Effect of Sulfur Compounds on Oxide Scale Formed in the Presence of Palmitic Acid. In order to assess the effect of sulfur compounds on the protectiveness of the oxide scale formed in the presence of PA, the CS specimens were pretreated in the autoclave and then challenged in the HVR. As shown in Table 7, solutions at TAN 1.75 were either a PA only solution (used as a baseline), or one of the two sulfur containing solutions: PA + Yellow oil or PA + DDS. In the pretreatment step, it was found that the corrosion rates of specimens exposed to PA + DDS solution were much higher than those exposed to a PA + Yellow oil, suggesting that DDS is more corrosive than the native sulfur compounds in

pretreatment solution

TAN 1.75 pretreatment corrosion rate (mm/y)

TAN 3.5 challenge corrosion rate (mm/y)

PA only %S = 0 wt % PA + DDS %S = 0.25 wt % PA + Yellow oil %S = 0.25 wt % bare steel TAN 3.5 corrosion ratea

0.2 0.7

4.8 4.2

0.4

1.8 7.8

a

Freshly polished specimens (no surface scale) were installed in the HVR and corroded by the challenge solution under corrosion challenge conditions.

Yellow oil. In both cases the corrosion rates were higher than those measured for the PA only solution. In the TAN 3.5 corrosion challenge step, the scale formed in PA only proved to be least protective, followed by the scale formed in the PA + DDS solution, with the scale formed in the PA + Yellow oil being most protective, as shown in Table 7. To put these corrosion rate numbers into context, the bare steel

Figure 2. Cross-sectional SEM images of CS specimens pretreated with PA only of TAN 0.5 (a), TAN 1 (b), and TAN 1.75 (c). (a′−c′) Results of corresponding EDS analyses performed along the white line. D

DOI: 10.1021/acs.energyfuels.6b01066 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

Figure 3. Cross-section SEM images of CS specimens pretreated with (a) PA only, (b) PA + DDS, or (c) PA + Yellow oil. (a′−c′) Results of corresponding EDS analysis performed along the white line.

Figure 4. Cross-sectional SEM images of CS rings after pretreatment and TAN 3.5 corrosion challenge: (a) PA only, (b) PA + DDS, and (c) PA + Yellow oil. (a′−c′) Results of corresponding EDS analysis performed along the white line.

The scale originally present on specimens pretreated with PA only was almost entirely dissolved, which is consistent with lack of protectiveness (Figure 4a). The lamellar structure seen on the surface is a remnant of a grain of pearlite (solid solution of α-ferrite and cementite Fe3C, a characteristic of austenitic steel). Note that there is a thin sulfide/oxide layer on the outer surface of the pearlite. Similar thin, outer sulfide and oxide layers have been seen with TCI in HVR parameter studies in the development of the corrosion challenge conditions.21 In those studies, nonprotective oxide layers were observed to grow with temperature, time and TCI concentration. Thus, the scales formed in PA only solutions adhere poorly and offer limited resistance to naphthenic acids. The scale formed in the PA + DDS solution survived the corrosion challenge but shows evidence of delamination as well as undermining which explains the poor protection this scale offered during the corrosion challenge (Figure 4b). On the other hand, the scale formed in PA + Yellow oil remained adherent after the corrosion challenge (Figure 4c). EDS analysis shows oxygen peaks on both sides of a tight sulfide layer. 3.3. Effect of Sulfur Compounds on Oxide Scale Formed in TCI. As a baseline for this series of experiments the pretreatment was done in TCI only solution at TAN 1.75 (see

corrosion rate obtained in a TAN 3.5 corrosion experiment (without sulfur compounds) is also shown there. The scales formed during the pretreatment step on specimen exposed to the sulfur containing PA + DDS and PA + Yellow oil solutions (shown in Figure 3b and c) are much thicker than that formed in the sulfur-free PA only solution (shown in Figure 3a). With the presence of DDS, the scale is thickest (around 7 μm) and composed of cracked layers that appear to contain predominantly iron and sulfur, as shown by the EDS analysis (Figure 3b). Oxygen is also detected but appears to be distributed within the iron sulfide layers. The scale formed on the specimen in the PA + Yellow oil solution is much thinner (around 3 μm), less granular, and the peak of oxygen appears closer to the steel surface. The challenge corrosion rates for the pretreated specimens reveal differences in protectiveness among the scales (Table 7). The scale formed in PA only reduces the challenge corrosion rate by about one-third. Although PA + DDS pretreatment generated a thicker scale, the protection appears to be similar to PA only solution. On the other hand, the scale formed in PA + Yellow oil is most protective. The appearance of the different scales remaining on the surface after the corrosion challenge is shown in SEM images in Figure 4. E

DOI: 10.1021/acs.energyfuels.6b01066 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

The pretreatment in TCI + Yellow oil lowered the challenge corrosion rate, but the TCI + DDS pretreatment had little effect. Both PA and TCI gave similar response in Yellow oil, but they behaved differently with or without DDS in Tufflo. Therefore, following analyses focus on the difference of scales formed in PA and TCI. 3.4. Morphology and Composition of Oxide Scale Formed in Acidic Solutions. SEM analyses show the presence of thin oxide scales on both PA only and TCI only pretreated rings (Figures 2 and 5). The PA scale appears to be ∼2 μm thick with an outer layer richer in carbon and the inner oxygen layer closer to the metal (Figure 2c). In contrast, the scale formed in TCI only appears to be thicker (>2 μm) consisting of both an outer ∼1 μm sulfur/oxygen layer and an inner ∼1 μm layer that appears to be pearlitic (Figure 5). However, the SEM resolution is not high enough to reveal the detailed scale morphology. Therefore, segment of scale from each were extracted by FIB and examined by high resolution TEM and EDS. The image for the PA only pretreatment scale shows a distribution of small (